Cant Believe the Difference

My host uses Webalizer, I looked at same log in Webtrends and...

[opulyst]

I couldn't believe the difference in the traffic statistics between Webalizer and Webtrends.

I mean my traffic in Webtrends is about 1/3 the traffic in the Webalizer.

Why do they read the log files so differently?

[mivox]

Do you have access to the settings/preferences file for both of them? One may be ignoring requests of a certain type that the other one is counting... (ie- requests for gifs, jpgs, requests from your own domain or internally referred requests...)

[opulyst]

I don't think I do. But I am not sure.

All I see in my log file are the daily log files zipped up.

Is there somewhere where I would see that? And is there a common name for the file.

[mivox]

Webalizer's default is "webalizer.conf" but if it's installed server-wide, you probably won't have access. Ask your host about it.

The last default installation of Webalizer I saw counted all the graphic file requests as hits (unless you're running an image download site, that's a waste)... and the "top referrers" list included internal pages from my site. Web Trands may be set to ignore those sorts of hits.

[oilman]

another thing to check is: Are the two programs using the same criteria to define a session? If you have one program saying a session is 30 minutes and one saying 45 min - the program saying 30 min will return higher traffic numbers.

Also check bandwidth usage and see how close it is in the two programs.

[ROLAND_F]

You will need to make a good configuration with webalizer or else you will record many bots as visitors. Hopefukky you can configure everything in Webalizer, you can tell him to ignore some kind of request and you will gain in accuracy.

[opulyst]

Webalizer is installed server wide. I will check into customizing the settings file, but I doubt that will get anywhere.

On the other hand Webtrends I have on my computer. I download the files from my site and have configured a report from the Webtrends software.

Is there something I can do to customize the Webtrends?

I was more concerned with the visitors, as hits are pretty much meaningless except for advertising and press releases.

Yeah, I think the images are being counted, they are always my top files each month.

I didn't realize about the session. Do you know where I could change this is in Webtrends?

I didn't realize the bots were counted. That makes sense though.

[ROLAND_F]

You can download Webalizer for Win32 and run it on your PC.
Download your logfile, configure webalizer (it's easy) run webalizer
and you will get accurate stats and nice html reports that are easy
to import in excell and you can also request text reports that you can parse in your own software.

Webalizer helped me to get the list of keywords used to reach my website for more than 12 month and now I have what I need to optimize
my webpages.

[webdiversity]

I'm undertaking evaluation of differet log analysis software and have noticed the discrepancies you mention. Sadly none of the software is proving accurate enough, some are mis-interpreting search engine traffic.

For example virtually all of Alta Vista traffic is not being picked up by FastStats, it also misses traffic from MSN and as the last poster mentioned that information of exactly how people got to your site in the first place is so important.

The server side offerings are so expensive that it virtually eliminates them as an option for smaller businesses.

[opulyst]

Oilman, I checked the bandwidth between the two, and Webtrends is about 1/10th of Webalizer.

Is there a way to customize Webtrends for session length, I have the software, but am not very savy at using it.

[Mark_A]

> I'm undertaking evaluation of differet log analysis software and have noticed the discrepancies you mention. Sadly none of the software is proving accurate enough, some are mis-interpreting search engine traffic.

If you can make your log files include all requests and get your log files into a database of your own you can identify the SE robots and screen / filter them out of visitor reports. When I used to do this they always asked for robots.txt first or at least sometime from the root and you could then isolate them by their IP.

They are useful info though because you get to see how often and how deep they visit I would prefer them always to be in the log.. they are after all a visit and server load, just not a human one.